ResearchBenchmarksPhysics Reasoning on Physics Puzzles Catapult (test)Follow44AccuracyHExA6.5616.282635.72Jun 28, 2026Evaluation ResultsMethodMethodLinksAccuracyImprovementHExALevels of Skills=DTE,...Levels of Skills=DTE, TBP, PTP, Solver=Sonnet2026.064436REACTSolver=SonnetSolver=Sonnet2026.068—